Research map / Point clouds / Segmentation and detection
Efficient Outdoor 3D Point Cloud Semantic Segmentation for Critical Road Objects and Distributed Contexts
Chi-Chong Wong, Chi-Man Vong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ADConv instead of Set Abstraction layer (ablation) | ADConvnet-SAGC · S3DIS Area-5 | mIoU: 57.4 -> 60.1 | Improved |
| add ADConv instead of Set Abstraction layer (ablation) | ADConvnet-SAGC · NPM3D Lille2 | mIoU: 77.5 -> 80.4 | Improved |
| add SAGC global context module (ablation) | ADConvnet-SAGC · S3DIS Area-5 | mIoU: 56.7 -> 60.1 | Improved |
| add SAGC global context module (ablation) | ADConvnet-SAGC · NPM3D Lille2 | mIoU: 76.3 -> 80.4 | Improved |
| use learned attention weights instead of constant weights (ablation) | ADConvnet-SAGC · S3DIS Area-5 | mIoU: 53.0 -> 60.1 | Improved |
| use learned attention weights instead of constant weights (ablation) | ADConvnet-SAGC · NPM3D Lille2 | mIoU: 71.6 -> 80.4 | Improved |
| add both ADConv and SAGC to a Set Abstraction baseline (ablation) | ADConvnet-SAGC · S3DIS Area-5 | mIoU: 52.8 -> 60.1 | Improved |
| add both ADConv and SAGC to a Set Abstraction baseline (ablation) | ADConvnet-SAGC · NPM3D Lille2 | mIoU: 72.2 -> 80.4 | Improved |
| increase neighbors K from 16 to 64 (ablation) | ADConvnet-SAGC · S3DIS Area-5 | mIoU: 59.0 -> 60.5 | Improved |
| increase neighbors K from 64 to 128 (ablation) | ADConvnet-SAGC · S3DIS Area-5 | mIoU: 60.5 -> 60.6 | No clear effect |
| use ADConvnet-SAGC instead of PointNet++ for inference | PointNet++ · NPM3D Lille2 | inference time (seconds): 756.2 -> 318.0 seconds | Improved |
Closest papers by meaning
- Image-to-Lidar Self-Supervised Distillation for Autonomous Driving Data (CVPR 2022 · similarity 0.89)
- Using a Waffle Iron for Automotive Point Cloud Semantic Segmentation (ICCV 2023 · similarity 0.89)
- Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR-Based Perception (TPAMI 2021 · similarity 0.89)
- SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences (ICCV 2019 · similarity 0.89)
- Attentional ShapeContextNet for Point Cloud Recognition (CVPR 2018 · similarity 0.89)
- VLM2Scene: Self-Supervised Image-Text-LiDAR Learning with Foundation Models for Autonomous Driving Scene Understanding (AAAI 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
